An evidence update on hypertension identified the following key points from new evidence:
1) Automated clinic blood pressure measurement providing multiple readings may be more accurate than manual measurement, but can still overestimate blood pressure compared to ambulatory monitoring.
2) Differences over 15 mmHg between arms on repeated measurements may indicate increased risk of vascular disease and mortality.
3) The long-term benefits of treating stage 1 hypertension without other risk factors remains uncertain.
The evidence update provided commentaries on these points and other new evidence related to diagnosing and managing hypertension. It indicated some evidence may impact current guidance, but overall the new evidence did not require changes to the guidance.
3. Contents
Introduction ................................................................................................................................ 4
Key points .................................................................................................................................. 5
1 Commentary on new evidence................................................................................... 7
1.1 Measuring blood pressure .......................................................................................... 7
1.2 Diagnosing hypertension ............................................................................................ 7
1.3 Assessing cardiovascular risk and target organ damage .......................................... 9
1.4 Lifestyle interventions ................................................................................................. 9
1.5 Initiating and monitoring antihypertensive drug treatment, including blood
pressure targets ......................................................................................................... 9
1.6 Choosing antihypertensive drug treatment .............................................................. 12
1.7 Patient education and adherence to treatment ........................................................ 21
Areas not currently covered by NICE guidance .................................................................. 22
2 New evidence uncertainties ..................................................................................... 24
Appendix A: Methodology ........................................................................................................ 25
Appendix B: The Evidence Update Advisory Group and Evidence Update project team ....... 27
Evidence Update 32 – Hypertension (March 2013) 3
4. Introduction
This Evidence Update identifies new evidence that is relevant to, and may have a potential
impact on, the following reference guidance:
1
Hypertension. NICE clinical guideline 127 (2011).
A search was conducted for new evidence from 29 November 2010 to 21 September 2012. A
total of 1948 pieces of evidence were initially identified. Following removal of duplicates and a
series of automated and manual sifts, 20 items were selected for the Evidence Update (see
Appendix A for details of the evidence search and selection process). An Evidence Update
Advisory Group, comprising topic experts, reviewed the prioritised evidence and provided a
commentary.
Although the process of updating NICE guidance is distinct from the process of an Evidence
Update, the relevant NICE guidance development centres have been made aware of the new
evidence, which will be considered when guidance is reviewed.
Feedback
If you have any comments you would like to make on this Evidence Update, please email
contactus@evidence.nhs.uk
1
NICE-accredited guidance is denoted by the Accreditation Mark
Evidence Update 32 – Hypertension (March 2013) 4
5. Key points
The following table summarises what the Evidence Update Advisory Group (EUAG) decided
were the key points for this Evidence Update. It also indicates the EUAG’s opinion on whether
the new evidence may have a potential impact on the current guidance listed in the
introduction. For further details of the evidence behind these key points, please see the full
commentaries.
The section headings used in the table below are taken from the guidance.
Evidence Updates do not replace current accredited guidance and do not provide
formal practice recommendations.
Potential impact
on guidance
Key point Yes No
Diagnosing hypertension
• Automated clinic blood pressure measurement providing a series
of readings may result in more accurate blood pressure
measurement than manual blood pressure measurement, but may
still overestimate blood pressure in some patients when compared
with ambulatory blood pressure monitoring.
• Differences of more than 15 mmHg in blood pressure readings
between arms may indicate increased risk of underlying vascular
disease and an increased risk of all cause and cardiovascular
mortality.
Initiating and monitoring antihypertensive drug treatment
including blood pressure targets
• The longer term benefit of drug treatment of stage 1 hypertension
remains uncertain in patients without target organ damage,
existing cardiovascular disease or at low risk of cardiovascular
disease.
• Telemonitoring of blood pressure control may result in greater
reductions in blood pressure when compared with usual care;
however further research is needed.
• Home blood pressure monitoring may be associated with greater
reductions in blood pressure compared with clinic blood pressure
monitoring; however, the effect size seems to be small and further
research is needed.
• There is uncertainty about the optimal timing of antihypertensive
medication and whether evening dosing versus morning dosing of
antihypertensive drugs is associated with any significant impact on
blood pressure reduction, cardiovascular outcomes or adverse
events – further research is needed.
Evidence Update 32 – Hypertension (March 2013) 5
6. Potential impact
on guidance
Key point Yes No
Choosing antihypertensive drug treatment
• Early treatment of hypertension may prolong life expectancy with
regard to cardiovascular mortality.
• Angiotensin-converting enzyme (ACE) inhibitors or angiotensin-II
receptor blockers (ARBs) seem to be associated with lower all-
cause and cardiovascular mortality when compared with control
(active control, placebo or usual care).
• Diuretics may be more effective than other classes of
antihypertensive drugs in preventing heart failure.
• Initial therapy with low-dose hydrochlorothiazide does not seem to
reduce blood pressure as much as other classes of
antihypertensive drugs.
• Initial antihypertensive therapy with beta-blockers may be less
effective than treatment with calcium channel blockers, diuretic
therapy and renin-angiotensin system blockers at preventing
cardiovascular events and mortality in people with hypertension.
Beta blockers may also be associated with greater
discontinuations due to adverse events than other classes of
antihypertensive drugs.
•
2
Spironolactone reduces blood pressure when compared with
placebo in people with resistant hypertension despite treatment
with at least 3 antihypertensive drugs.
• Limited evidence suggests that loop diuretics may reduce blood
pressure compared with placebo; more research is needed.
• The maximum blood pressure lowering effect of antihypertensive
treatment may be estimated from the response seen in the first
1–2 weeks of treatment with an antihypertensive drug.
Patient education and adherence to treatment
• People with hypertension appear to have similar perspectives on
hypertension and their treatment, and the reasons for non-
adherence to therapy do not seem to vary substantially between
geographical and ethnic groups.
Areas not currently covered by NICE guidance
• An implantable device designed to activate baroreceptors to
reduce blood pressure does not appear to reduce blood pressure
compared with control in people with uncontrolled hypertension.
Further studies are needed to determine safety and efficacy.
• A guided breathing device does not appear to be an effective
treatment to reduce blood pressure in people with hypertension.
2
At the time of publication of this Evidence Update, spironolactone did not have UK marketing
authorisation for this indication.
Evidence Update 32 – Hypertension (March 2013) 6
7. 1 Commentary on new evidence
These commentaries analyse the key references identified specifically for the Evidence
Update. The commentaries focus on the ‘key references’ (those identified through the search
process and prioritised by the EUAG for inclusion in the Evidence Update), which are
identified in bold text. Supporting references provide context or additional information to the
commentary. Section headings are taken from the guidance.
1.1 Measuring blood pressure
No new key evidence was found for this section.
1.2 Diagnosing hypertension
Automatic versus manual measurement and correlation with ambulatory blood
pressure monitoring (ABPM)
NICE CG127 recommends offering ABPM to confirm the diagnosis of hypertension in people
with a clinic blood pressure of 140/90 mmHg or higher.
Myers et al. (2011) reported results from the conventional versus automated measurement of
blood pressure in the office (CAMBO) randomised controlled trial (RCT), in which 88
Canadian family doctors who used manual sphygmomanometers were randomly assigned by
site (67 practices) to use automated clinic blood pressure measurement with the BpTRU
automated sphygmomanometer or to continue with manual blood pressure measurement.
The primary outcome measure was the difference between awake systolic ABPM and
automated clinic blood pressure compared with the difference between awake systolic ABPM
and manual clinic blood pressure.
The study included patients with hypertension who were aged older than 45 years with at
least 160 mmHg systolic and less than 95 mmHg diastolic for untreated people and at least
140 mmHg systolic and less than 90 mmHg diastolic for people receiving antihypertensive
drugs. Exclusion criteria were serious coexisting illness or diabetes, more than twice the
normal value of serum creatinine, or a history of non-adherence to therapy. All participants
(n=555) had 24 hour ABPM after the clinic measurement. The size of groups differed (n=303
in the automated measurement group and n=252 in the manual measurement group) partly
because more of the sites randomly assigned to automated measurement had multiple
clinicians working in the practice.
The BpTRU device took an initial measurement to verify the correct positioning of the cuff,
then the participant was left alone while a further 5 readings were taken at 2-minute intervals.
No additional instructions on proper blood pressure measurement technique were given to
doctors in either the manual or automated measurement groups.
The mean difference in blood pressure between automated clinic measurement and awake
average ABPM was −2.3 mmHg (95% confidence interval [CI] −0.3 to −4.3, p=0.02) for
systolic pressure and −3.3 mmHg (95% CI −2.2 to −4.4, p<0.001) for diastolic pressure. The
mean difference in blood pressure between manual clinic blood pressure measurement and
ABPM was −6.5 mmHg (95% CI −4.3 to −8.6, p<0.001) for systolic pressure and −4.3 mmHg
(95% CI −2.9 to −5.8, p<0.001) for diastolic pressure. For systolic blood pressure, the
difference between the automated group and the manual group was significant (p=0.006), but
the difference in diastolic blood pressure was not significant.
The authors noted that further research is needed before these results can be applied to
hypertensive populations other than that studied in this trial.
Evidence Update 32 – Hypertension (March 2013) 7
8. This evidence suggests that automated clinic blood pressure measurement based on multiple
readings may give a more accurate blood pressure reading when compared with manual
clinic blood pressure measurement. However, automatic clinic blood pressure measurement
still resulted in a significantly higher reading than the corresponding ABPM awake average
blood pressure. These results support the recommendation in NICE CG127 to confirm a clinic
diagnosis of hypertension with ABPM.
Key reference
Myers MG, Godwin M, Dawes M et al. (2011) Conventional versus automated measurement of blood
pressure in primary care patients with systolic hypertension: randomised parallel design controlled trial.
BMJ 342: d286
Differences in blood pressure between arms
NICE CG127 recommends measuring blood pressure in both arms when considering a
diagnosis of hypertension. If the difference in readings between arms is more than 20 mmHg
then the measurements should be repeated. If the difference in readings between arms
remains more than 20 mmHg on the second measurement, subsequent measurements
should be conducted in the arm with the higher reading. The full version of NICE CG127
notes that such large differences between arms indicates underlying vascular disease and
that consistent inter-arm differences of more than 20/10 mmHg may suggest pathology
warranting specialist referral.
Clark et al. (2012) conducted a meta-analysis of 20 cohort or cross-sectional studies of
differences in systolic blood pressure between arms in populations of adults of 18 years or
older with data for central vascular disease, peripheral vascular disease or death.
From 5 case series (n=135) of people with angiographically proven asymptomatic subclavian
stenosis mean systolic blood pressure was estimated to be 36.9 mmHg (95% CI 35.4 to 38.4)
lower in the affected arm than in the other. Pooling of a further 2 datasets (n=532) allowed
estimation of the risk of subclavian stenosis of more than 50% occlusion at angiography and a
difference of 10 mmHg or more between arms (relative risk [RR]=8.8, 95% CI 3.6 to 21.2,
p<0.0001).
There was no significant association between a difference of 15 mmHg or more between
arms and a history of coronary artery disease when compared with a difference between
arms of less than 15 mmHg in 7 cohorts (RR=1.13 95% CI 0.78 to 1.63, n=5033). A between-
arm difference of 15 mmHg or more was associated with significantly increased risks of all
cause (hazard ratio [HR] =1.55, 95% CI 1.07 to 2.25) and cardiovascular mortality (HR=1.68,
95% CI 1.11 to 2.53). The authors noted that they had insufficient data to establish an effect
of small studies or publication bias.
The results of this study suggest a difference in systolic blood pressure of more than
15 mmHg between arms may indicate increased risk of vascular disease (that is, subclavian
stenosis), and is associated with an increased risk of all cause and cardiovascular mortality.
These results provide some support for the existing NICE CG127 recommendation to:
measure blood pressure in both arms and use the arm with the highest blood pressure
measurement for all subsequent readings. It is also noted in full version of NICE CG127 that
differences in blood pressure between arms of 20 mmHg or more may also indicate
underlying pathology that warrants specialist investigation.
Key reference
Clark CE, Taylor RS, Shore AC et al. (2012) Association of a difference in systolic blood pressure
between arms with vascular disease and mortality: a systematic review and meta-analysis. The Lancet:
DOI:10.1016/S0140-6736(11)61710-8
Evidence Update 32 – Hypertension (March 2013) 8
9. 1.3 Assessing cardiovascular risk and target organ damage
No new key evidence was found for this section.
1.4 Lifestyle interventions
New evidence was not searched for this section (see Appendix A for details of the evidence
search and selection process).
1.5 Initiating and monitoring antihypertensive drug treatment,
including blood pressure targets
Drug treatment for mild hypertension
NICE CG127 recommends antihypertensive drug treatment for people aged under 80 years
with stage 1 hypertension (clinic blood pressure 140/90 to 160/100 mmHg and subsequent
ABPM daytime average or home blood pressure monitoring [HBPM] average of 135/85 to
150/95 mmHg) and one or more of: target organ damage, established cardiovascular disease,
renal disease, diabetes, or 10-year cardiovascular risk of 20% or higher.
In a Cochrane review, Diao et al. (2012), assessed 4 trials (n=8912) of antihypertensive drug
treatment in people with mild hypertension (blood pressure of 140–159 mmHg systolic or 90–
99 mmHg diastolic) and no evidence of cardiovascular disease at baseline.
After 4–5 years of treatment with antihypertensive drugs, no significant differences were seen
for total mortality (RR=0.85, 95% CI 0.63 to 1.15) coronary heart disease (n=7080, RR=1.12,
95% CI 0.80 to 1.57), stroke (RR=0.51, 95% CI 0.24 to 1.08), or total cardiovascular events
(RR=0.97, 95% CI 0.72 to 1.32) compared with placebo. Data on withdrawals due to adverse
effects were only available from all patients in one of the included trials and not from the
subgroup of patients with mild hypertension. Withdrawals due to adverse events among
participants with mild to moderate hypertension (n=17,354) were more common in people on
antihypertensive drug treatment than placebo. (RR=4.80, 95% CI 4.14 to 5.57, absolute risk
increase 8.9% over 5 years).
This review is consistent with NICE CG127, in that evidence is lacking for benefits of
antihypertensive treatment for stage 1 hypertension in people who do not have coexisting
conditions that may increase their risk of cardiovascular events.
Key reference
Diao D, Wright JM, Cundiff DK et al. (2012) Pharmacotherapy for mild hypertension. Cochrane
Database of Systematic Reviews issue 8: CD006742
Telemonitoring
NICE CG127 does not include recommendations on telemonitoring; it states that the
response to antihypertensive treatment should be monitored by clinic blood pressure
measurements; ABPM or HBPM may be considered as an adjunct to clinic measurements for
people identified as having a ‘white-coat effect’.
Verberk et al. (2011) did a meta-analysis of 9 RCTs (n=2501) evaluating telemonitoring of
hypertension that reported change in blood pressure or proportion of participants reaching
their target blood pressure.
Overall, telemonitoring decreased systolic blood pressure significantly more than usual care
(mean difference of change with telemonitoring – change with usual care=5.2 mmHg standard
deviation [SD]±1.5, p<0.001). Diastolic blood pressure also reduced significantly more with
telemonitoring than with usual care (mean difference=2.1 SD ±0.8 mmHg, p<0.01. Additional
Evidence Update 32 – Hypertension (March 2013) 9
10. analysis to account for possible publication bias showed slightly smaller differences which
were still significant (data not reported).
Although no tests for heterogeneity were reported, the authors noted that the included studies
differed in equipment used, participants, blood pressure endpoints and study design. The
authors did not comment on the quality of included studies.
The results of the study by Verberk et al. (2011) suggest there may be a small reduction in
blood pressure in people with treated hypertension who use telemonitoring of their blood
pressure control. However, further research is needed. This evidence is insufficient to affect
current recommendations in NICE CG127, that is, to continue to monitor blood pressure
control using clinic blood pressure unless the patient has been previously determined to have
a ‘white-coat effect’, in which case blood pressure control should be monitored with home
blood pressure readings or ABPM.
Key reference
Verberk WJ, Kessels AG, Thien T (2011) Telecare is a valuable tool for hypertension management, a
systematic review and meta-analysis. Blood Pressure Monitoring 16: 149–55
Home versus clinic blood pressure monitoring
NICE CG127 recommends monitoring response to blood pressure lowering treatment by
measuring blood pressure in the clinic. However, ABPM or HBPM may be considered as an
adjunct to clinic measurements in people with an identified white-coat effect.
A meta-analysis by Agarwal et al. (2011) assessed HBPM compared with clinic blood
pressure measurement in 37 RCTs (n=9446) to determine whether HBPM resulted in
improvements in blood pressure control or more responsive changes in antihypertensive drug
use.
The mean change in systolic blood pressure was −2.63 mmHg (95% CI −4.24 to − 1.02,
p<0.0001) for those using HBPM when compared with clinic measurement to monitor blood
pressure control: the mean change in diastolic blood pressure was −1.68 mmHg (95% CI
−2.58 to −0.79, p<0.0001). Subgroup analysis showed that studies that did not use drug
titration based on the results of HBPM resulted in lower blood pressure (mean=−2.46, 95% CI
−3.44 to −1.49) than studies that did use drug titration (mean=−0.58, 95% CI −1.94 to 0.79,
p=0.028). However, significantly greater response rates were seen with drug titration.
The authors reported significant heterogeneity between studies in both systolic and diastolic
blood pressure and in response rates for HBPM. They additionally noted the paradox in that
mean blood pressure was lowered more, but that response rates increased when no drug
titration was used.
In a single-centre RCT, Fuchs et al. (2012) compared HBPM and no changes to drug
regimens with usual care in patients with uncontrolled hypertension (n=136). Participants
were randomly assigned to 1 of 4 groups: HBPM, pharmacist care, HBPM plus pharmacist
care, and usual care. This report pooled the results of the two HBPM groups as the
intervention group and pooled the pharmacist only and usual care groups as the control
group. Exclusion criteria included blood pressure of 180/110 mmHg or higher and serious
coexisting conditions. The primary outcomes were the differences between 24 hour, mean
night-time and mean day-time blood pressure measured by ABPM between baseline and the
end of the trial.
Baseline blood pressure was the mean of 6 measurements in 3 visits taken with an aneroid
sphygmomanometer and the mean of 4 measurements in 2 visits with an automatic
oscillometric blood pressure measuring device. Clinic hypertension was defined as
140/90 mmHg or higher. All participants additionally had 24 hour ABPM, for which
hypertension was defined as 130/80 mmHg or higher.
Evidence Update 32 – Hypertension (March 2013) 10
11. People assigned to HBPM used an automatic oscillometric device and a blood pressure diary
and were instructed to take 6 measurements every day. At follow up visits at 7, 30, and
60 days the readings from the automatic oscillometric device were correlated with the blood
pressure diary. At 7 days, instructions were reinforced; at subsequent follow-up visits patients
were given advice on lifestyle modifications to help with lowering blood pressure.
People assigned to usual care had follow-up at 7, 30 and 60 days to receive the same advice
as the intervention group. 24 hour ABPM was repeated in all participants at the end of the
study and analysis was done only for 121 participants who completed this second round of
ABPM. Pharmacist care was not defined.
Adherence to HBPM was 94% at the end of the first week and 85% at the end of the trial. The
change in ABPM results between baseline and end of trial in the HBPM group was
significantly more than that of the control group for both systolic blood pressure (mean
[SD]=8.8±13.1 mmHg vs 3.4±11.6 mmHg respectively, p=0.02) and diastolic blood pressure
(mean [SD]=5.6±8.4 mmHg vs 1.0±7.9 mmHg respectively, p=0.002). More people in the
HBPM group had blood pressure within target at the end of the study (32.4%) compared with
control (16.2%, p=0.03).
The authors noted that their results showed a larger effect than two recent meta-analyses in
this area, one of which was the Agarwal et al. (2011) meta-analysis. They postulated that this
difference in results probably resulted from differences in the protocol of this study compared
with studies included in the meta-analyses. Author-noted limitations included that more than
60% of participants were women, which could affect the applicability of findings to men, and
the short follow-up.
This new evidence is not likely to affect NICE CG127 because the RCT reported by Fuchs et
al. (2012) was small, and in the meta-analysis by Agarwal et al. (2011) the effect size was
small and similar to that found in the evidence review in the full version of NICE CG127.
Further research into HBPM is needed.
Key references
Agarwal R, Bills JE, Hecht TJ et al. (2011) Role of home blood pressure monitoring in overcoming
therapeutic inertia and improving hypertension control: a systematic review and meta-analysis.
Hypertension 57: 29–38
Fuchs SC, Ferreira-da-silva AL, Moreira LB et al. (2012) Efficacy of isolated home blood pressure
monitoring for blood pressure control: randomized controlled trial with ambulatory blood pressure
monitoring – MONITOR study. Journal of Hypertension 30: 75–80
Morning versus evening dosing
Recommendations in NICE CG127 do not specify the time of day antihypertensive drug
treatment should be taken.
A Cochrane review by Zhao et al. (2011) assessed 21 RCTs (n=1993) lasting at least 3
weeks that compared dosing a single antihypertensive drug (angiotensin-converting enzyme
[ACE] inhibitors, alpha blockers, angiotensin-II receptor blockers [ARBs], beta-blockers,
calcium channel blockers [CCBs], or diuretics) dosed in the morning (6 am to 12 noon) with
evening dosing (6 pm to 12 midnight) of the same drug. Participants had primary
hypertension (systolic blood pressure 140 mmHg or higher).
The authors stated that the primary outcomes of interest were death from all causes,
cardiovascular mortality and cardiovascular morbidity; however, none of the included trials
reported any of these outcomes. All trials reported the change from baseline in 24 hour blood
pressure.
The mean differences in systolic (−1.71 mmHg, 95% CI −2.78 to −0.65 mmHg) and diastolic
(−1.38 mmHg, 95% CI −2.13 to −0.62 mmHg) blood pressure were significantly lower for the
Evidence Update 32 – Hypertension (March 2013) 11
12. evening regimen. In 5 trials the rates of overall adverse events (RR=0.78 95% CI 0.37 to
1.65) and withdrawals due to adverse events (RR=0.53, 95% CI 0.26 to 1.07) did not differ
significantly between morning and evening regimens.
Most included studies were assessed as at high risk of bias due to lack of allocation
concealment or selective reporting and there was significant heterogeneity. The authors noted
that the current evidence suggests that taking antihypertensive drugs in the evening may
reduce 24 hour blood pressure; however, data were not available to determine which regimen
has beneficial effects on cardiovascular outcomes or adverse events. This evidence is not
likely to have an impact on the recommendations in NICE CG127 and further research is
needed.
Key reference
Zhao P, Xu P, Wan C et al. (2011) Evening versus morning dosing regimen drug therapy for
hypertension. Cochrane Database of Systematic Reviews issue 10: CD004184
1.6 Choosing antihypertensive drug treatment
Treating hypertension in older people
NICE CG127 recommends the same drugs for people 80 years and over as people aged
55–80 years. However, the target blood pressure for people aged 80 years and over is
150/90 mmHg for clinic measurements, and 145/85 mmHg when using ABPM and HBPM (for
example, in people identified as having a 'white coat effect' and people who choose to monitor
their blood pressure at home).
Beckett et al. (2012) reported on a 1-year open-label extension trial of the Hypertension in
the Very Elderly Trial (HYVET). Eligible participants were those in HYVET who had not had a
primary or secondary endpoint event by the end of the double-blind phase (n=1882), and
1712 of them consented to take part. The data from the 1-year extension were considered
separately from the results of the randomised trial, and compared outcomes in people who
were on active treatment in the double-blind phase (n=924) with those of people on placebo
during the double-blind phase (n=788). Participants were at least 80 years of age (mean
85 years). To be eligible for the main trial, participants had to have had systolic blood
pressure greater than 160 mmHg over a 2-month run-in period.
The main HYVET study (Beckett et al. 2008) was stopped early, after only a median of
1.8 years of follow-up, because of clear evidence of benefit from active treatment: total
mortality was reduced by 21% in relative terms (95% confidence interval 4% to 35%;
p=0.019). Additionally, intention-to-treat analyses showed that active treatment was
associated with significant reductions in the rates of death from stroke (39%, 95% CI 1% to
62%, p=0.05), and rates of heart failure (64%, 95% CI 42% to 78%, p<0.001). No significant
differences were seen for rates of fatal and non-fatal stroke (30% reduction, 95% CI −1% to
51%, p=0.06) or death from cardiovascular causes (23% reduction, 95% CI −1% to 40%,
p=0.06)
At the start of the open-label extension (Beckett et al. 2012), mean sitting blood pressure was
145.0/76.6 mmHg in the people previously on active treatment, and 159.3/80.8 mmHg in
those previously on placebo (p<0.001). The treatment steps in the main trial were:
indapamide modified release 1.5 mg; indapamide plus perindopril 2 mg; and indapamide plus
perindopril 4 mg. All participants were started on indapamide at the start of the extension
study irrespective of whether they were taking indapamide or indapamide plus perindopril at
the end of the main trial. The use of additional antihypertensive drugs was allowed if the
target was not reached in the extension study, but only 0.9% of people were taking these at
6 months and only 1.8% at the end of the extension study.
Evidence Update 32 – Hypertension (March 2013) 12
13. The difference between active and placebo groups at the end of the double blind phase was
14.3/4.2 mmHg (p<0.001); however by 6 months the difference was reduced to 1.3/0.6 mmHg
(not statistically significant, p value not reported) and at the end of the extension study the
difference was 1.0/0.2 mmHg.
Fatal and non-fatal stroke rates (the primary endpoint) did not differ significantly between the
groups previously on active or placebo treatment (5.18 versus 9.89 events per 1000 patient
years, HR=1.92, 95%CI 0.59 to 6.22, p=0.28). No significant differences were seen in heart
failure (HR=0.28, 95% CI 0.03 to 2.73, p=0.28) or cardiovascular events (HR=0.78, 95% CI
0.36 to 1.72, p=0.55). The rates of total mortality (HR=0.48, 95% CI 0.26 to 0.87, p=0.02) and
cardiovascular mortality (HR=0.19, 95% CI 0.04 to 0.87, p=0.03) were significantly lower in
those who had active treatment in the double-blind phase. Adjustment for age, sex, baseline
sitting systolic blood pressure, and previous cardiovascular disease did not change the
results.
The main limitations discussed by the authors were the short duration of follow-up and small
numbers of events, and that only people still on double-blind treatment at the end of the
original trial were included in the extension study. Any person who had an endpoint event in
the main trial had been moved to open label treatment and was not eligible to enter the
extension study. This study was an extension to an RCT and a relatively low number of
events were recorded, so it is probably underpowered to reach any definitive conclusions.
The results suggest that the benefits on cardiovascular morbidity and mortality in those
previously allocated to active antihypertensive treatment in the main trial may have continued
into this extension study. The HYVET study was influential in the development of
recommendations for the treatment of people aged 80 years and over as covered in the
evidence review in the full version of NICE CG127. This extension study provides
supplemental evidence in support of these recommendations.
Key reference
Beckett N, Peters R, Tuomilehto J et al. (2012) Immediate and late benefits of treating very elderly
people with hypertension: results from active treatment extension to Hypertension in the very elderly
randomised controlled trial. BMJ 344: d7541
Supporting reference
Beckett NS, Peters R, Fletcher AE et al. (2008) Treatment of hypertension in patients 80 years of age or
older. New England Journal of Medicine 358: 1887–98
Long-term mortality benefits of antihypertensive treatment
NICE CG127 recommends offering antihypertensive drug treatment to people of any age with
stage 2 hypertension. Additionally, people with isolated systolic hypertension (systolic blood
pressure 160 mmHg or more) should be offered the same treatment as people with both
raised systolic and diastolic blood pressure, and a thiazide-like diuretic should be offered if
diuretic treatment is to be initiated or changed.
Kostis et al. (2011) reported on 22-year follow up of an RCT (Systolic Hypertension in the
Elderly Program, SHEP) of chlortalidone in people older than 60 years. At baseline,
participants (n=4736) had a systolic blood pressure of 160–219 mmHg and diastolic blood
pressure of less than 90 mmHg (mean 170.3/76.6 mmHg) and their mean age was
71.6 years. They were randomised to active treatment or placebo: active treatment started
with chlortalidone 12.5 mg, increasing to 25 mg if necessary, with the option of adding
atenolol if blood pressure targets were not reached. The target was a systolic blood pressure
of less than 160 mmHg for those with a systolic blood pressure of more than 180 mmHg at
entry, and a reduction in systolic blood pressure of at least 20 mmHg in the other participants.
All participants were advised to take active treatment at the end of the trial. Long-term follow-
Evidence Update 32 – Hypertension (March 2013) 13
14. up was done by matching personal identifiers of participants to the US National Death Index
until 31 December 2006.
At the end of the RCT (4.5 years), the mean systolic blood pressure was 26 mmHg lower than
baseline in the active treatment group and 15 mmHg lower in the placebo group. Mean
diastolic blood pressure was 9 mmHg lower than at baseline for the active treatment group
and 4–5 mmHg lower in the placebo group. Active treatment led to a statistically significant
decrease in the risk of fatal and non-fatal stroke (primary endpoint, RR=0.64, 95% CI 0.50 to
0.82), myocardial infarction (RR=0.67, 95% CI 0.47 to 0.96) and heart failure (RR=0.51, 95%
CI 0.37 to 0.71). No statistically significant differences were noted in the risk of death from
cardiovascular causes (OR=0.80, 95% CI 0.60 to 1.05) or all-cause mortality (OR=0.87, 95%
CI 0.73 to 1.05).
During the 22 year follow up, total mortality did not differ significantly between the people who
had been on active treatment in SHEP (59.9%) and those who had been on placebo (60.5%,
p=0.38). Similar numbers of people in both groups died from coronary heart disease (12.4%
vs 13.6% respectively) and stroke (4.6% vs 5.6% respectively): statistical analysis was not
reported. However, people who had been on active treatment in the SHEP RCT had
significantly longer life expectancy before cardiovascular death, with a gain of 158 days in 22
years (95% CI 36 to 287 days, p=0.009), increasing to 215 days (95% CI 70 to 346 days, p
value not reported) in those whose blood pressure was controlled to SHEP targets by the end
of the RCT. Life expectancy for all-cause death was not significantly different (p=0.07).
The authors reported that their study was limited by the fact that treatments beyond the
randomised phase were observational and information on treatments for hypertension and
other conditions such as diabetes, hyperlipidaemia, and surgical and device interventions was
not available. They additionally noted that the confidence intervals for life expectancy gain
were very wide.
The results of this study suggest that early antihypertensive treatment may lead to
prolongation of life expectancy with regard to cardiovascular mortality. This benefit was
observed despite the fact that the original clinical trial comparing active versus placebo
treatment lasted only 4.5 years and that during the subsequent follow-up of 22 years, all
participants were offered antihypertensive therapy. These data provide evidence to support:
the longer term benefits of early antihypertensive therapy; the treatment of isolated systolic
hypertension; and treatment with thiazide-like diuretics. Therefore this evidence is broadly
consistent with recommendations in NICE CG127.
Key reference
Kostis JB, Cabrera J, Cheng JQ et al. (2011) Association between chlorthalidone treatment of systolic
hypertension and long-term survival. JAMA 306: 2588–93
ACE-inhibitors or ARBs
NICE CG127 recommends ACE inhibitors or low-cost ARBs for step 1 treatment for people
over 55 years. If an ACE inhibitor is prescribed and is not tolerated (for example, because of
cough), a low-cost ARB should be offered. For people younger than 55 years and black
African or Caribbean origin of any age NICE CG127 recommends CCBs as initial therapy.
In a meta-analysis, van Vark et al. (2012) included 20 randomised controlled trials of at least
100 patients (n=159,998) comparing ACE inhibitors or ARBs with control (placebo, active
control or usual care) with an incidence of death of at least 10 people and follow-up of at least
1 year. Studies in which less than two-thirds of participants met each trial’s definition of
hypertension were excluded.
The analyses were based on the mortality incidence rate: that is, the number of participants
who died divided by the number of patient-years of follow-up. The overall mean follow-up time
Evidence Update 32 – Hypertension (March 2013) 14
15. was 4.3 years. ACE inhibitors were the active treatment in 7 trials (n=76,615) and 13 trials
used ARBs as active treatment (n=82,383). The mean baseline systolic blood pressure was
153 mmHg (range of means 135–182 mmHg) and the mean age was 67 years (range of
means 59–84 years).
People in the active treatment groups (ACE inhibitors or ARBs) had an all-cause mortality rate
of 20.9 deaths per 1000 patient years compared with 23.3 deaths per 1000 patient years in
the control group. The hazard ratio for all-cause mortality was significantly lower for active
treatment compared with control (HR=0.95, 95% CI 0.91 to 1.00, p=0.032), with no significant
heterogeneity between trials. Cardiovascular death was 8.7 deaths per 1000 patient-years for
active treatment versus 10.1 deaths per 1000 patient-years for control (HR=0.93, 95% CI 0.88
to 0.99, p=0.018).
The effect on all-cause mortality was almost completely due to the effect of ACE inhibitors,
which were associated with a 10% reduction in all-cause mortality compared with control
(HR=0.90, 95% CI 0.84 to 0.97, p=0.004). However, ARBs were not associated with a
reduction in overall mortality (HR=0.99, 95% CI 0.94 to 1.04, p=0.683). The difference
between ACE inhibitors and ARBs was significant (p=0.036). ACE inhibitors were associated
with a reduction in death from cardiovascular causes which was of borderline statistical
significance (HR 0.88, 95% CI 0.77 to 1,00, p=0.051). The effect of ARBs on this outcome
was not significant (HR 0.96, 95% CI 0.90 to 1.01, p=0.143). Meta-regression analysis
showed a significant association between mean baseline systolic blood pressure and
reductions in mortality (p=0.035). The mortality reduction was highest in trials with the highest
baseline systolic blood pressure. Additionally, the mortality reduction was largest in the trials
with larger reductions in mean systolic blood pressure compared with control. No significant
relationships with mortality were seen for age, sex, or mean follow-up time.
Author-reported limitations included variations in study populations, for example, in definitions
of hypertension, dosages of drugs used, or target blood pressure. The meta-analysis was
based on trial-level data rather than individual patient data and was not designed to make a
head-to-head comparison between ACE inhibitors and ARBs: the authors stated that the
difference in effects on all-cause mortality should be considered a post-hoc observation, and
should not affect current clinical practice. The inclusion of trials comparing multiple-drug
regimens is a potential confounding factor because the effects on blood pressure cannot be
reliably attributed to one drug or class of drugs.
This meta-analysis revealed marginal treatment effects on mortality and interpretation is
complex due to the heterogeneity between studies and because these effects may be directly
related to the overall blood pressure reduction rather than to an effect of a particular class of
drugs. This analysis does not influence the recommendations in NICE CG127 to offer either
ACE inhibitors or low-cost ARBs as step 1 antihypertensive treatment when indicated.
Key reference
van Vark LC, Bertrand M, Akkerhuis KM et al. (2012) Angiotensin-converting enzyme inhibitors reduce
mortality in hypertension: a meta-analysis of randomized clinical trials of renin–angiotensin–aldosterone
system inhibitors involving 158 998 patients. European Heart Journal 33: 2088–97
Antihypertensive drugs in prevention of heart failure
NICE CG127 recommends offering ACE inhibitors or ARBs as preferred first line treatment for
people younger than 55 years. CCBs are recommended for people older than 55 years and
black people of African or Caribbean origin of any age. Step 2 treatment recommends an
ACE inhibitor or ARB plus a CCB. If a CCB is not tolerated or the person has oedema,
evidence of heart failure, or high risk of heart failure, a thiazide-like diuretic should be
considered.
Evidence Update 32 – Hypertension (March 2013) 15
16. In a Bayesian network meta-analysis of 26 randomised controlled trials (n=223,313),
Sciarretta et al. (2011) assessed the efficacy of antihypertensive drugs in preventing heart
failure. Included studies were published from 1997 to 2009 because the clinical features of
patients in trials in that period would be similar to current clinical practice. Inclusion criteria
were: studies of at least 200 people; a population with diagnosed hypertension or at high
cardiovascular risk with 65% or more of participants with hypertension; and information on the
absolute incidence of heart failure and other major cardiovascular events.
Analysis was done for classes of antihypertensive drugs: ACE inhibitors; alpha-blockers;
ARBs; beta-blockers; conventional treatment; CCBs; and diuretics compared with placebo
and with each other. Conventional treatment varied by study and was beta-blockers or
diuretics in 4 studies and any drug other than ARBs in 2 studies (mostly CCBs).
Heart failure occurred in 8554 people (3.8%) over all studies. Compared with placebo, the
most effective drugs for preventing heart failure were diuretics (odds ratio [OR]=0.59, 95%
credibility interval [CrI] 0.47 to 0.72), then ACE inhibitors (OR=0.71, 95% CrI 0.58 to 0.84),
then ARBs (OR=0.76, 95% CrI 0.62 to 0.90) followed by CCBs (OR=0.83, 95% CrI 0.67 to
0.99). Beta-blockers (OR 0.87, 95% CrI 0.64 to 1.12) and alpha-blockers (OR=1.22, 95% CrI
0.85 to 1.69) did not have a significant effect on development of heart failure. In the network
meta-analysis, diuretics were more effective than CCBs (OR=0.71, 95% CrI 0.60 to 0.86) and
ACE inhibitors (OR=0.83, 95% CrI 0.69 to 0.99). Beta-blockers (OR=1.47, 95% CrI 1.10 to
1.92) and ARBs (OR=1.28, 95% CrI 1.04 to 1.59) were less effective than diuretics.
The authors noted that variations in the criteria for assessing heart failure and the doses of
drugs used varied between trials, which may have affected the heart failure incidence seen in
the included studies. Another potential limitation noted by the authors was that analyses were
generally of first-line treatment, so multiple-drug regimens could not be considered. The
analysis also did not take account of the effect of differences in blood pressure reduction
obtained with different therapies compared with placebo. Additionally, the authors discussed
that fact that this study only addresses the effect of antihypertensive drugs on heart failure,
but that other studies have found benefits of specific classes on other important
cardiovascular outcomes such as myocardial infarction or stroke.
The recommendation in NICE CG127 to use CCB as the preferred initial or step 2 therapy for
people aged 55 years and over, contained an important caveat, that is, the sensitivity analysis
in the full version of NICE CG127 had suggested that thiazide-like diuretics would be the
preferred treatment in patients with heart failure or at high risk of developing heart failure. The
evidence from this recent analysis showing that diuretics may be the most effective treatment
to prevent heart failure lends support to the recommendations in NICE CG127.
Key reference
Sciarretta S, Palano F, Tocci G et al. (2011) Antihypertensive treatment and development of heart
failure in hypertension. A Bayesian network meta-analysis of studies in patients with hypertension and
high cardiovascular risk. Archives of Internal Medicine 171: 384–94
Step 3 treatment
Diuretics
NICE CG127 recommends adding a thiazide-like diuretic at step 3 of the antihypertensive
treatment algorithm. If diuretic treatment is to be initiated or changed, a thiazide-like diuretic
should be offered, such as chlortalidone (12.5–25.0 mg once daily) or indapamide (1.5 mg
modified-release once daily or 2.5 mg once daily) in preference to a conventional thiazide
diuretic such as bendroflumethiazide or hydrochlorothiazide. For people who are already
having treatment with bendroflumethiazide or hydrochlorothiazide and whose blood pressure
Evidence Update 32 – Hypertension (March 2013) 16
17. is stable and well controlled, treatment with bendroflumethiazide or hydrochlorothiazide
should be continued.
Messerli et al. (2011) did a meta-analysis of 19 trials (n=1463) of hydrochlorothiazide
monotherapy compared with active therapy that evaluated efficacy by ABPM. The included
studies assessed 12.5–25 mg (14 trials, n=1234) or 50 mg doses (5 trials, n=229). The
primary outcome was reduction in blood pressure from baseline to follow-up assessed by
24-hour ABPM.
After a mean of 17 weeks of treatment, hydrochlorothiazide 12.5–25 mg resulted in a
reduction in blood pressure from baseline of −6.5 mmHg (95% CI −5.3 to −7.7 mm Hg)
systolic and −4.5 mmHg (95% CI −3.1 to −6.0 mmHg) diastolic. The 12.5 mg and 25 mg
doses separated did not give significantly different results. Other antihypertensive drugs (ACE
inhibitors, ARBs, beta-blockers and CCBs resulted in greater reductions in blood pressure
(p<0.001). Mean absolute differences were 4.5–6.2/2.9–6.7 mmHg.
For the hydrochlorothiazide 50 mg group, the reductions were −12.0 mmHg (95% CI −8.2 to
−15.9 mmHg) systolic and −5.4 mmHg (95% CI −3.2 to −7.7 mmHg): the systolic reduction
was statistically significantly better than the 25 mg dose (p=0.04) but the diastolic reduction
was not (p=0.97). However, the authors noted that adverse events would limit the use of
doses higher than 25 mg daily.
Of the 14 trials of hydrochlorothiazide 12.5–25 mg, 4 were assessed as low risk of bias, the
rest were at high risk of bias. Heterogeneity was also reported to be an issue.
An open-label, 2-year RCT reported by Trimarco et al. 2012 enrolled 2409 people starting
antihypertensive drugs, who were randomly assigned to chlortalidone 12.5–25 mg (n=1205)
or to any other class of drugs other than diuretics (control group, n=1204). If blood pressure
targets were not met, any other antihypertensive including diuretics could be added. People
aged 18–75 years (mean 54 years) with previously untreated or poorly controlled stage 1 or 2
hypertension were included. People with clinically significant coexisting conditions were
excluded.
Participants had clinic blood pressure measurements taken monthly by aneroid
sphygmomanometer until clinic blood pressure was within target range (140/90 mmHg or
130/80 mmHg for people with diabetes), thereafter it was measured every 2 months. Each
participant started treatment with 1 drug, which was titrated to the highest dose before
another was added based on the GP’s judgement. The primary outcome was adherence to
treatment.
The study was completed by 87% of those in the chlortalidone group and 85% of those in the
control group. No significant difference was seen in adherence scores between the
chlortalidone and control groups (0.91 vs 0.91). Significantly more people in the chlortalidone
group had modified treatment (79%) compared with those not on first-line diuretics (44%,
p<0.0001). Inadequate control of blood pressure was the most common reason for treatment
modification (58% of modifications in the chlortalidone group and 37% in the control group,
p<0.0001). In addition to the open-label nature of the study, a potential limitation of the study
noted by the authors was that no pill count was done to confirm concordance with treatment.
The study by Messerli et al. (2011) demonstrates that low-dose hydrochlorothiazide may not
be as effective as other antihypertensive treatments at lowering blood pressure. The findings
of this analysis are consistent with NICE CG127, which recommends thiazide-like diuretics
(for example, chlortalidone or indapamide) should be preferred to thiazide diuretics
(hydrochlorothiazide or bendroflumethiazide) when diuretic treatment is indicated for the
treatment of hypertension. The study by Trimarco et al. (2012) provides support for the
recommendations in NICE CG127, which noted that diuretics may not be the optimal first-line
antihypertensive treatment for many people.
Evidence Update 32 – Hypertension (March 2013) 17
18. Key references
Messerli FH, Makani H, Benjo A et al. (2011) Antihypertensive efficacy of hydrochlorothiazide as
evaluated by ambulatory blood pressure monitoring. Journal of the American College of Cardiology 57:
590–600
Trimarco V, de Simone G, Izzo R et al. (2012) Persistence and adherence to antihypertensive treatment
in relation to initial prescription: diuretics versus other classes of antihypertensive drugs. Journal of
Hypertension 30: 1225–32
Beta-blockers for hypertension
NICE CG127 does not recommend beta-blockers as initial routine therapy for uncomplicated
hypertension. Beta-blockers may be considered as step-1 treatment in younger people who
have intolerance or contraindication to ACE inhibitors or ARBs, women of child-bearing
potential, or people with evidence of increased sympathetic drive.
Wiysonge et al. (2012) conducted a Cochrane review of 13 RCTs (n=91,561) of beta-
blockers compared with placebo, no treatment or active controls for the first-line treatment of
hypertension. The primary outcome was total mortality.
Total mortality did not differ significantly between beta-blockers and placebo (4 trials,
n=23,613, RR=0.99, 95% CI 0.88 to 1.11); beta-blockers and diuretics (5 trials, n=18,241,
RR=1.04, 95% 0.91 to 1.19); or beta-blockers and ACE inhibitors or ARBs (3 trials, n=10,828,
RR=1.10, 95% CI 0.98 to 1.24). However, it was significantly higher for beta-blockers
compared with CCBs (4 trials, n=44,825, RR=1.07, 95% 1.00 to 1.14), which corresponded to
an absolute risk increase of 0.5% over 5 years.
There was no evidence that beta blockers reduced the risk of coronary heart disease when
compared with placebo; ACE inhibitors or ARBs; CCBs; or diuretics. The risk of fatal or non-
fatal stroke was significantly lower for beta-blockers compared with placebo (4 trials,
n=23,613, RR=0.80, 95% CI 0.66 to 0.96). However, stroke reduction with beta blockers was
less than that seen with CCBs (3 trials with 44,167 participants, RR 1.24, 95% CI 1.11 to
1.40) and ACE-inhibitor or ARBs (2 trials with 9951 participants, RR 1.30, 95% CI 1.11 to
1.53), but no significant difference was seen for diuretics (4 trials with 18,135 participants,
RR=1.17, 95% CI 0.65 to 2.09).
Discontinuations due to adverse events did not differ significantly between beta-blockers and
placebo, diuretics or CCBs. People on a beta-blocker had significantly higher rate of
discontinuation than those on ACE inhibitors or ARBs (2 trials, n=9951, RR=1.41, 95% CI
1.29 to 1.54).
The authors stated that this analysis of 13 RCTs suggests that first-line beta-blockers for
elevated blood pressure were not as good at decreasing mortality and morbidity as other
classes of drugs: diuretics, calcium channel blockers, and renin-angiotensin system inhibitors.
The authors concluded that initiating treatment of hypertension with beta-blockers may lead to
modest reductions in cardiovascular disease without significant effects on mortality. These
effects of beta-blockers seem to be inferior to those of other antihypertensive drugs. This
conclusion is consistent with current guidance in NICE CG127 that does not consider beta
blockers to be the most effective initial treatment for hypertension.
Key reference
Wiysonge CS, Bradley HA, Volmink J et al. (2012) Beta-blockers for hypertension. Cochrane Database
of Systematic Reviews issue 8: CD002003
Evidence Update 32 – Hypertension (March 2013) 18
19. Spironolactone in resistant hypertension
NICE CG127 recommends considering a low dose of spironolactone (25 mg daily) as an
option for adding to the antihypertensive regimen at step 4 of the treatment algorithm if blood
potassium level is 4.5 mmol/l or lower. At the time of publication of this Evidence Update,
spironolactone did not have UK marketing authorisation for this indication.
Václavík et al. (2011) reported results of the addition of spironolactone (25 mg) in patients
with resistant arterial hypertension (ASPIRANT) study. This randomised placebo-controlled,
double blind trial enrolled 117 adults with resistant hypertension defined as clinic blood
pressure greater than 140 mmHg systolic or 90 mmHg diastolic despite treatment with at least
3 antihypertensive drugs including a diuretic. People with diabetes or chronic kidney disease
were enrolled if their blood pressure exceeded 130 mmHg systolic or 80 mmHg diastolic. The
mean age of participants was about 61 years and body-mass index was about 32 kg/m². The
mean number of antihypertensive drugs was 4.6 for people in the spironolactone group and
4.5 in the placebo group. Mean clinic blood pressure was 154/92 mmHg and 24 hour ABPM
was 141/80 mmHg. Mean daytime ABPM was 142/82 mmHg.
The following groups were excluded: people with severe hypertension (greater than
180 mmHg systolic or 110 mmHg diastolic) needing immediate adjustment of treatment, renal
insufficiency (GFR <40 ml/min); pregnant or lactating women or women of childbearing
potential not using effective contraception;, hyperkalaemia greater than 5.4 mmol/l or
hyponatraemia less than 130 mmol/l; people with known hypersensitivity to spironolactone;
and people already taking an aldosterone antagonist. Participants had clinic visits at 4 and
8 weeks, with an additional visit at 2 weeks scheduled for people older than 75 years, and
those with diabetes or chronic kidney disease. At each visit, blood pressure was measured
3 times with a calibrated mercury sphygmomanometer and the mean of the second and third
readings was recorded. 24-hour ABPM was conducted at baseline and week 8.
The trial was stopped early at the first interim analysis because of significant reductions in
systolic blood pressure. However, diastolic blood pressure did not reduce as much as
expected and a revised power calculation estimated that about 5 times more participants
would be needed to detect a significant difference in diastolic blood pressure between groups.
Spironolactone was associated with a greater reduction in ABPM daytime systolic blood
pressure than placebo (mean difference=−5.4 mmHg, 95% CI −10.0 to −0.8 mmHg, p=0.024).
The difference in diastolic blood pressure was not significant (mean difference=−1.0 mmHg,
95% CI −4.0 to 2.0 mmHg, p=0.358). The target clinic systolic blood pressure of less than
140 mmHg at 8 weeks was reached in 30 patients (54.5%) using spironolactone and in 24
patients (42.9%) using placebo (p=0.257). Adverse events were reported to be similar
between groups with no significant difference in severe adverse events leading to treatment
discontinuation (2 events for spironolactone and 1 event for placebo, p=0.618). Serum
potassium increased by a median 0.3 mmol/l in the spironolactone group.
The authors noted that the lack of a significant difference in diastolic blood pressure may be
influenced by the relatively low baseline diastolic blood pressure (mean office reading
92 mmHg) and that 38% of participants had isolated systolic hypertension.
The results of this study are consistent with the recommendation in NICE CG127 to consider
adding spironolactone to the antihypertensive regimen at step 4 of the treatment algorithm.
Nevertheless, the number of studies of the optimal treatment for resistant hypertension
identified in the full version of NICE CG127 was limited, and further research is needed.
Key reference
Václavík J, Sedlák R, Plachý M et al. (2011) Addition of spironolactone in patients with resistant
hypertension (ASPIRANT): a randomized, double-blind, placebo-controlled trial. Hypertension 57: 1069–
75
Evidence Update 32 – Hypertension (March 2013) 19
20. Loop diuretics
Loop diuretics are not recommended in the antihypertensive treatment algorithm in NICE
CG127.
Musini et al. (2012) did a Cochrane review of 9 double-blind placebo-controlled trials (n=460)
of loop diuretics with at least 3 weeks of follow-up. Included studies had to recruit participants
(mean age 54.4 years) with blood pressure of 140 mmHg or more systolic or 90 mmHg or
more diastolic and could not titrate dose according to blood pressure response. This
Cochrane review was an update of a previous publication, and the search period for new data
was January 2009 to February 2012; no new trials were identified.
The best estimate of blood pressure reduction was −7.9 mmHg systolic (95% CI−10.5 to
−5.4 mmHg) and −4.4 mmHg diastolic (95% CI−5.6 to −2.8 mmHg). There were no significant
differences between diuretics and placebo in withdrawals due to adverse events.
Generally, the included studies were noted to be at high risk of bias in terms of selective
reporting and at low risk of bias in terms of blinding. The authors noted that the meta-analysis
did not provide a good estimate of the incidence of harm associated with loop diuretics
because of the short duration of included studies. The authors concluded that more RCTs are
needed to determine the effects of loop diuretics.
NICE CG 127 does not recommend the use of loop diuretics for the routine treatment of
hypertension because, as indicated by this Cochrane review, the evidence base for blood-
pressure lowering with loop diuretics remains limited and the studies have mainly been small
and of short duration. This new analysis is not likely to affect the recommendations in NICE
CG127.
Key reference
Musini VM, Rezapour P, Wright JM et al. (2012) Blood pressure lowering efficacy of loop diuretics for
primary hypertension. Cochrane Database of Systematic Reviews issue 8: CD003825
Timing of stepping up treatment
NICE CG127 does not contain recommendations on what timescales should be used when
considering stepping up antihypertensive treatment.
Lasserson et al. (2011) systematically reviewed 18 studies (n=4168) including at least 20
participants in each arm who started antihypertensive treatment after a placebo washout
period and reported change from baseline in clinic systolic and diastolic blood pressures at
least every 2 weeks after starting treatment. The mean reductions in systolic and diastolic
blood pressures were extracted and the data were used to derive parameters for a model of
the effect of time on blood pressure response.
All included studies used diastolic blood pressure for inclusion so no participants had isolated
systolic hypertension. The half-lives of study drugs were 2–24 hours and effects of dose were
assessed either by titration during treatment phases or by random assignment to fixed doses.
The instrument used for measuring blood pressure and the number of readings taken differed
between studies.
On average and across all drugs and doses, the time taken for people to reach 50% of the
maximum predicted response to antihypertensive drugs appeared to be about 1 week, and
most of the blood-pressure lowering appeared to occur within 4 weeks. When analysis
included the effects of a titration schedule, the time to 50% maximum response was longer
(1.2 weeks vs 0.7 weeks for systolic and 1.4 weeks vs 0.7 weeks for diastolic), and the
plateau was higher (p values not reported).
The data did not show particular benefits of any class of antihypertensive drugs over another;
however, the authors noted that there may not have been sufficient power to detect
Evidence Update 32 – Hypertension (March 2013) 20
21. differences between classes. They suggested that estimation of maximum response could be
done after 1–2 weeks to help clinicians to determine when a newly started antihypertensive
drug can be judged to be ineffective. Additionally, after 4 weeks little further benefit can be
expected. However, further research is needed to see whether treatment strategies based on
these results would result in better outcomes for patients.
This evidence suggests that healthcare professionals need not wait longer than 4 weeks
between titration of antihypertensive drugs. It is not likely to affect recommendations in NICE
CG127, but may be helpful to health professionals and patients when deciding when to move
to the next treatment step if blood pressure is not controlled.
Key reference
Lasserson DS, Buclin T, Glasziou P (2011) How quickly should we titrate antihypertensive medication?
Systematic review modelling blood pressure response from trial data. Heart 97: 1771–5
1.7 Patient education and adherence to treatment
Lay perspectives of hypertension and treatment adherence
NICE CG127 includes recommendations to enhance patient education and adherence to
treatment, and recognises that people vary in their attitudes to their hypertension and their
experience of treatment. Providing details of patient organisations is highlighted as potentially
being helpful. Interventions to overcome practical problems associated with non-adherence
are recommended if a specific need is identified because evidence supporting interventions to
increase adherence is inconclusive.
Marshall et al. (2012) did a systematic review of 53 qualitative studies based on face-to-face
interviews and focus groups about patients’ perspectives of hypertension and drug treatment
in people with uncomplicated hypertension that were published in peer-reviewed journals
(number of participants not reported). Studies based on telephone interviews or
questionnaires were excluded, as were studies with more than 50% of participants with
cardiovascular disease, diabetes or who were pregnant.
Many studies (20) were from the USA, 8 were from the UK, 7 were from Brazil, and the
remainder were from other countries worldwide; 24 studies included only participants from
minority ethnic groups. Topics covered were: patients’ understanding of causes, effects,
exacerbating factors and consequences of hypertension; attitudes to drug treatments; and the
perceived influences of stress, diet and racism.
Major themes of the research included that people thought that hypertension was related to
stress or worrying about work or family life or experiencing racism. Some people felt that their
hypertension was related to a particular stressful event in their lives. Hypertension was also
thought by some people to be temporary or curable, and to be distinct from high blood
pressure by others.
Most people recognised that hypertension caused serious complications such as stroke or
heart disease. Some people reported that taking drugs reduced anxiety or worries mostly
thought to be a direct effect of the drugs, but a few people thought that reduction in worry was
due to protection from the complications of hypertension. Participants in 2 studies thought that
antihypertensive drugs functioned as sedatives.
Headache and dizziness were commonly reported as possible symptoms of hypertension, but
many people reported no symptoms of hypertension. Some people reported taking drugs
according to the prescription, but many reported deliberate non-adherence, often because
people took the drug according to their perceived symptoms or level of stress, or to avoid
side-effects of the drugs. Unintentional non-adherence was due to forgetfulness, busy lives,
and healthcare costs.
Evidence Update 32 – Hypertension (March 2013) 21
22. The quality of each study was scored, and the score was used as one indicator of the
robustness of the findings. Studies with a low quality score were not excluded because of lack
of agreement by qualitative researchers on criteria to use, many possible qualitative methods
and the role of subjective judgement in analysis.
Contrary to the conclusions of individual studies suggesting that culture-specific education is
needed, perceptions of hypertension were similar across ethnic and geographical
populations. People from minority ethnic groups reported experiencing racism in many
studies, the stress of which was thought to exacerbate hypertension. Additionally, migrant
populations perceived that they were more likely to have low paying jobs and more likely to
have economic hardship.
The authors noted that to overcome problems with non-adherence, clinicians and educational
interventions need to acknowledge and incorporate patients’ concerns and perspectives. For
example, instead of denying the possibility of symptoms, it may be more helpful to say that
many people report symptoms but they are not a reliable indication of fluctuations in blood
pressure.
Although this study suggests that future educational interventions should incorporate and
engage with commonly held beliefs about hypertension and adherence to treatment, it may
also be useful for clinicians to discuss these topics with their patients. This evidence is not
likely to affect recommendations in NICE CG127.
Key reference
Marshall IJ, Wolfe CD, McKevitt C (2012) Lay perspectives on hypertension and drug adherence:
systematic review of qualitative research. BMJ 344: e3953
Areas not currently covered by NICE guidance
Devices for lowering blood pressure
Baroreflex activation therapy
NICE CG127 does not recommend any device-based methods for the treatment of
hypertension.
Bisognano et al. (2011) reported on an RCT of baroreflex activation therapy (BAT) in people
with uncontrolled hypertension (clinic blood pressure 160/80 mmHg or higher and systolic
APBM 135 mmHg or more). All participants had the Rheos system implanted by a vascular,
cardiothoracic or neuro-surgeon. The Rheos system consisted of a pulse generator with two
leads, one of which was attached to each carotid sinus. Participants (n=265) were randomly
assigned 2:1 to have the device turned on 1 month after the implantation procedure or
6 months later.
The trial had 5 primary endpoints: acute efficacy, sustained efficacy, procedural safety, BAT
safety, and device safety. All endpoints needed to be met to demonstrate overall safety and
efficacy. Efficacy endpoints were based on a reduction in systolic blood pressure of 10 mmHg
or more. At baseline, people were taking an average of 5.2 (SD 1.7) antihypertensive drugs.
At 6 months, response (defined as the proportion of subjects that achieved at least a
10 mmHg reduction in systolic blood pressure compared with baseline, and with a superiority
margin of 20%), was not significantly different in the group with devices switched on (54%)
compared with those whose devices remained off (46%, p=0.97). The group with devices
switched on showed a mean reduction in blood pressure of 7 mmHg compared with those
with devices switched off (p=0.08). 88% of those who responded at 6 months maintained
response at 12 months (p=0.001).
Evidence Update 32 – Hypertension (March 2013) 22
23. The event-free surgery rate was 74.8%, which was lower than the prespecified safety level of
82% (p=1.00); BAT safety was 91.7% in those whose device was turned on after 1 month and
89.3% in those whose device was switched on 6 months later (p<0.001). Device safety was
reported as an event-free rate of 87.2%, which was higher than the prespecified safety level
of 72% (p<0.001).
This paper suggests modest blood pressure lowering with this device; however, the authors
noted that they did not meet 2 of the 5 endpoints; that further trials are needed; and that a
less invasive implant procedure has been developed, which may increase safety in future
surgical procedures. This technology is unlikely to influence recommendations in NICE
CG127.
Key reference
Bisognano JD, Bakris G Nardim MK et al. (2011) Baroreflex activation therapy lowers blood pressure in
patients with resistant hypertension: results from the double-blind, randomized placebo-controlled
Rheos pivotal trial. Journal of the American College of Cardiology 58: 765–73
Device-guided breathing
NICE CG127 does not recommend any device-based methods for the treatment of
hypertension.
In a meta-analysis, Mahtani et al. (2012) assessed 8 trials (n=494) of a device designed to
lower breathing rate (Resperate) with the aim of reducing blood pressure. The device uses a
breathing monitor and plays tones via headphones to guide inhalation and exhalation, aiming
for less than 10 breaths per minute. Control interventions included meditation exercise,
relaxing music, blood pressure monitoring or usual care.
Overall, device-guided breathing resulted in a reduction in systolic (−3.06 mmHg, 95% CI
−4.68 to −1.43, p=0.0002) and diastolic blood pressure (−2.35 mmHg, 95% CI −3.47 to −1.22,
p=0.0001). However, the risk of bias was assessed as moderate in 6 studies and high in 2
others. The authors noted concern that 5 of the trials were conducted or funded by the
manufacturer of the device. Excluding these trials left 3 trials in 100 people, which showed no
overall effect on systolic (−1.97, 95% CI −6.50 to 2.56, p=0.39) or diastolic blood pressure
(−0.04, 95% CI −3.67 to 3.59).
All trials were short, with a maximum intervention duration of 9 weeks, and the authors noted
variable compliance across studies, so current data do not allow any conclusions to be made
about long-term efficacy. The authors concluded that the overall positive results of device-
guided breathing ‘should be interpreted with caution because of study size, cost of device,
variability in study quality and potential conflicts of interest from the trial sponsors and the
manufacturers of the Resperate device.’
Further independent trials are needed to assess the blood-pressure lowering efficacy of this
device. The available evidence does not seem to support the use of this device for the
treatment of hypertension and is not likely to influence the recommendations in NICE CG127.
Key reference
Mahtani KR, Nunan, D, Heneghan CJ (2012) Device-guided breathing exercises in the control of human
blood pressure: systematic review and meta-analysis. Journal of Hypertension 30: 852–60
Evidence Update 32 – Hypertension (March 2013) 23
24. 2 New evidence uncertainties
During the development of the Evidence Update, the following evidence uncertainties were
identified for the NHS Evidence UK Database of Uncertainties about the Effects of
Treatments (UK DUETs).
Initiating and monitoring antihypertensive drug treatment, including blood pressure
targets
• Evening versus morning dosing regimen drug therapy for hypertension
• Pharmacotherapy for mild hypertension
Choosing antihypertensive drug treatment
• Blood pressure lowering efficacy of loop diuretics for primary hypertension
• Different antihypertensive strategies in patients with diabetes, nephropathy and a history
of myocardial infarction for the treatment of hypertension and prevention of heart failure
Further evidence uncertainties for hypertension can be found in the UK DUETs database and
in the NICE research recommendations database.
UK DUETs was established to publish uncertainties about the effects of treatments
that cannot currently be answered by referring to reliable up-to-date systematic reviews of
existing research evidence.
Evidence Update 32 – Hypertension (March 2013) 24
25. Appendix A: Methodology
Scope
The scope of this Evidence Update is taken from the scope of the reference guidance:
• Hypertension. NICE clinical guideline 127 (2011).
The theme of lifestyle interventions was not covered because this body of literature is unlikely
to have changed in recent years; certain non-drug interventions were included; and a new
theme of blood pressure thresholds, targets or goals was included.
Searches
The literature was searched to identify studies and reviews relevant to the scope. Searches
were conducted of the following databases, covering the dates 29 November 2010 (the end of
the search period of NICE clinical guideline 127) to 21 September 2012:
• CDSR (Cochrane Database of Systematic Reviews)
• CINAHL (Cumulative Index to Nursing and Allied Health Literature)
• EMBASE (Excerpta Medica database)
• MEDLINE (Medical Literature Analysis and Retrieval System Online)
• NHS EED (Economic Evaluation Database)
Table 1 provides details of the MEDLINE search strategy used, which was adapted to search
the other databases listed above. The search strategy was based on the searches used for
NICE CG127 with some changes to reflect the changes in scope, and was used in
conjunction with validated Scottish Intercollegiate Guidelines Network search filters for RCTs
and systematic reviews.
One other study (Clark et al. 2012) was also identified outside of the literature search. Figure
1 provides details of the evidence selection process. The long list of evidence excluded after
review by the Chair of the EUAG, and the full search strategies, are available on request from
contactus@evidence.nhs.uk
There is more information about how NICE Evidence Updates are developed on the NHS
Evidence website.
Table 1 MEDLINE search strategy (adapted for individual databases)
1 exp *Hypertension/ ((elevat$ or high or increas$) adj blood
6 pressure?).ti.
2 (hypertens$ or prehypertens$).ti.
((elevat$ or high$ or increas$) adj
(hypertens$ or prehypertens$).ab.
(systolic or diastolic or arterial) adj
3 /freq=2
7 pressure?).ti.
((elevat$ or high or increas$) adj
8 or/1-7
(systolic or diastolic or arterial) adj
4 pressure?).ab. /freq=2 (pre eclampsia or preeclampsia or
pregnan$ or malignan$ or portal or
((elevat$ or high or increas$) adj blood
9 renal or ocular or intracranial).ti.
5 pressure?).ab. /freq=2
10 8 not 9
Evidence Update 32 – Hypertension (March 2013) 25
26. Figure 1 Flow chart of the evidence selection process
EUAG – Evidence Update Advisory Group
Evidence Update 32 – Hypertension (March 2013) 26
27. Appendix B: The Evidence Update Advisory
Group and Evidence Update project team
Evidence Update Advisory Group
The Evidence Update Advisory Group is a group of topic experts who review the prioritised
evidence obtained from the literature search and provide the commentary for the Evidence
Update.
Professor Bryan Williams – Chair
Professor of Medicine, Director; National Institute for Health Research University College
London Hospitals Biomedical Research Centre, Institute of Cardiovascular Science,
University College London.
Professor Mark Caulfield
Director, William Harvey Research Institute, Barts and the London School of Medicine and
Dentistry, Queen Mary University of London
Dr John Crimmins
General Practitioner, Vale of Glamorgan
Dr Terry McCormack
General Practitioner, Spring Vale Medical Centre, North Yorkshire
Professor Richard McManus
Professor of Primary Care, Department of Primary Care Health Sciences, University of Oxford
Michaela Watts
Hypertension Nurse Specialist, Addenbrooke's Hospital, Cambridge
Helen Williams
Consultant Pharmacist for Cardiovascular Disease, South London Cardiovascular and Stroke
Network
Evidence Update project team
Marion Spring
Associate Director
Chris Weiner
Consultant Clinical and Public Health Adviser
Cath White
Programme Manager
Lynne Kincaid
Medical Writer
Bazian
Information specialist support
Evidence Update 32 – Hypertension (March 2013) 27